One performance.
Every annotation, explained.

These student predictions approximate the original embedding teachers; they do not establish a person's actual identity.

{
  "orange_timbre_128d": [
    -0.04983147606253624,
    0.023882592096924782,
    -0.10610577464103699,
    -0.0997358113527298,
    0.008297921158373356,
    -0.09964712709188461,
    -0.03593664616346359,
    0.023651329800486565,
    0.06697720289230347,
    -0.0924522876739502,
    0.20571409165859222,
    -0.011516612954437733,
    0.05164948105812073,
    -0.08872520178556442,
    -0.07791733741760254,
    -0.06398153305053711,
    -0.1117740347981453,
    -0.07198496907949448,
    0.06820324063301086,
    -0.08438748866319656,
    0.16487917304039001,
    0.05484118312597275,
    -0.0926118865609169,
    -0.04861260578036308,
    -0.09429089725017548,
    -0.0910421684384346,
    -0.029078027233481407,
    -0.05176115781068802,
    -0.06215202808380127,
    0.12419476360082626,
    0.1089307963848114,
    0.0925208181142807,
    -0.0588122233748436,
    -0.04414619132876396,
    -0.10966892540454865,
    0.24869053065776825,
    -0.10426044464111328,
    0.01955111138522625,
    -0.08710084110498428,
    -0.10220997780561447,
    -0.09901589900255203,
    -0.11599769443273544,
    -0.06689763069152832,
    -0.10077015310525894,
    0.24835310876369476,
    -0.08648348599672318,
    -0.019590899348258972,
    -0.12266386300325394,
    -0.11799351125955582,
    0.0491182766854763,
    0.060477349907159805,
    -0.02687610127031803,
    0.10162234306335449,
    0.06204751878976822,
    -0.0650954470038414,
    0.14747367799282074,
    0.09298334270715714,
    -0.10431508719921112,
    -0.08620548993349075,
    0.028218740597367287,
    -0.07901781797409058,
    0.1483258754014969,
    -0.07442032545804977,
    0.01107813697308302,
    0.048053257167339325,
    -0.094272680580616,
    0.09073445945978165,
    -0.0988701805472374,
    0.10603292286396027,
    -0.09497438371181488,
    0.0841708555817604,
    0.039952028542757034,
    -0.09752141684293747,
    0.0005181399756111205,
    -0.021209031343460083,
    -0.019901476800441742,
    0.01118406467139721,
    0.08703230321407318,
    0.0013645407743752003,
    -0.0694628655910492,
    0.03915470093488693,
    0.08340346068143845,
    0.02051367424428463,
    -0.017722204327583313,
    0.05281371250748634,
    0.08200916647911072,
    0.14894993603229523,
    -0.009487091563642025,
    0.06415023654699326,
    0.016483550891280174,
    -0.11295408755540848,
    0.20560479164123535,
    -0.10499858111143112,
    -0.004711559042334557,
    -0.010504337027668953,
    0.0008507116581313312,
    -0.09244557470083237,
    -0.06340733170509338,
    0.03034818544983864,
    -0.02579406090080738,
    0.09905904531478882,
    0.07521548867225647,
    0.08731508255004883,
    -0.11588840186595917,
    -0.051474057137966156,
    -0.07182537019252777,
    -0.042012572288513184,
    0.03049173392355442,
    0.16161222755908966,
    -0.09537988156080246,
    0.026091240346431732,
    -0.10134435445070267,
    -0.1000320166349411,
    -0.012575422413647175,
    -0.11623924970626831,
    0.09887256473302841,
    -0.08654724061489105,
    0.03970590978860855,
    -0.03912954404950142,
    0.16081945598125458,
    -0.02512662298977375,
    -0.012262250296771526,
    -0.07197108119726181,
    -0.04720009118318558,
    0.03151121735572815,
    -0.021381694823503494,
    -0.08782985806465149,
    0.1260218769311905
  ],
  "orange_identity_250d": [
    -0.08017516136169434,
    -0.04697108268737793,
    -0.023682555183768272,
    -0.04240821674466133,
    0.02732660062611103,
    -0.05781247466802597,
    0.051901888102293015,
    0.08717372268438339,
    0.00547076016664505,
    0.05490074306726456,
    -0.03769021853804588,
    0.01832989789545536,
    -0.021137867122888565,
    -0.07172944396734238,
    -0.09888685494661331,
    0.10896465182304382,
    -0.060961294919252396,
    0.08402490615844727,
    -0.01852688379585743,
    -0.04651955887675285,
    0.03301751986145973,
    -0.05272827297449112,
    0.019229575991630554,
    0.030205167829990387,
    0.07397162169218063,
    -0.02379758469760418,
    -0.05233076214790344,
    0.06562010943889618,
    0.12665806710720062,
    -0.020517179742455482,
    0.04940852150321007,
    -0.03464426100254059,
    -0.03757864236831665,
    0.021381082013249397,
    0.03180048614740372,
    0.032644350081682205,
    0.14495831727981567,
    -0.024882107973098755,
    -0.0975164845585823,
    -0.03537474945187569,
    -0.033714957535266876,
    -0.032381124794483185,
    0.050255872309207916,
    -0.014267432503402233,
    0.030649779364466667,
    0.010242458432912827,
    -0.006765435449779034,
    -0.02773023582994938,
    0.032653119415044785,
    0.012189097702503204,
    -0.047851596027612686,
    -0.0012340941466391087,
    0.006127479951828718,
    -0.009498809464275837,
    0.03165580704808235,
    0.01514094602316618,
    -0.026761701330542564,
    -0.017197459936141968,
    -0.05620838701725006,
    0.08087608218193054,
    -0.05105797201395035,
    -0.04922724887728691,
    -0.06058121100068092,
    -0.04398611560463905,
    0.03971969336271286,
    0.05562954768538475,
    -0.017495619133114815,
    0.10652361065149307,
    -0.05982103943824768,
    0.02571728639304638,
    -0.07251051813364029,
    0.06497842818498611,
    -0.03167323023080826,
    0.05153911933302879,
    -0.056190963834524155,
    0.10473819822072983,
    0.08438756316900253,
    0.007102253381162882,
    -0.015668829903006554,
    0.09301820397377014,
    -0.05515177547931671,
    -0.03183533623814583,
    -0.027595054358243942,
    -0.06738798320293427,
    0.02115180715918541,
    -0.040852975100278854,
    -0.004098452627658844,
    0.07027184218168259,
    -0.017852095887064934,
    -0.005180888809263706,
    -0.06039987877011299,
    0.003340501105412841,
    -0.047403495758771896,
    -0.03688643127679825,
    -0.03700149059295654,
    -0.053056083619594574,
    -0.05178677290678024,
    -0.07945681363344193,
    0.13546644151210785,
    -0.020493559539318085,
    -0.004691308829933405,
    0.037861090153455734,
    -0.01941523514688015,
    0.018338583409786224,
    -0.023815065622329712,
    -0.10543563961982727,
    -0.04503227397799492,
    -0.07762259989976883,
    0.13431572914123535,
    -0.048526324331760406,
    -0.061330921947956085,
    0.0959298238158226,
    -0.031014179810881615,
    0.05313275381922722,
    -0.0719839558005333,
    -0.05459732934832573,
    -0.0352928526699543,
    0.04117381572723389,
    -0.0851721242070198,
    -0.005229719914495945,
    -0.08680060505867004,
    0.030464965850114822,
    0.007210718467831612,
    -0.022807270288467407,
    -0.003642191644757986,
    -0.044169239699840546,
    0.1343366503715515,
    -0.037768684327602386,
    -0.022599831223487854,
    0.021405477076768875,
    -0.050186172127723694,
    -0.04481957480311394,
    0.13360436260700226,
    -0.030876412987709045,
    0.10866472125053406,
    -0.04148414731025696,
    -0.036128003150224686,
    -0.05841571092605591,
    0.04962824285030365,
    0.12116242200136185,
    -0.04404192790389061,
    -0.02022683434188366,
    -0.048310089856386185,
    -0.06117050349712372,
    -0.07381816953420639,
    -0.058544766157865524,
    0.13888372480869293,
    -0.031188540160655975,
    0.008042979054152966,
    -0.017450258135795593,
    0.17051872611045837,
    0.0611182302236557,
    0.011935440823435783,
    -0.04694494605064392,
    0.0258637648075819,
    -0.05338040739297867,
    -0.04493812099099159,
    0.00251152366399765,
    -0.016679635271430016,
    -0.06427749991416931,
    -0.06282337754964828,
    -0.07139814645051956,
    -0.09689223021268845,
    -0.06830509006977081,
    -0.008947448804974556,
    -0.01614084467291832,
    -0.003454487770795822,
    0.049249958246946335,
    -0.014637933112680912,
    -0.03212312236428261,
    -0.04904940724372864,
    -0.045795924961566925,
    0.007000768557190895,
    0.09809184074401855,
    -0.06572813540697098,
    -0.07138072699308395,
    0.15009130537509918,
    -0.0591445118188858,
    -0.09300415217876434,
    -0.011687826365232468,
    0.1761537790298462,
    -0.031052516773343086,
    -0.037081699818372726,
    -0.010071570053696632,
    -0.05802517011761665,
    0.05669661611318588,
    0.007093995343893766,
    0.14679253101348877,
    0.03751586005091667,
    0.10104887932538986,
    0.05635138228535652,
    0.047030381858348846,
    -0.02637549117207527,
    -0.05404637008905411,
    0.01588364876806736,
    -0.024213530123233795,
    0.1647440493106842,
    0.0038148798048496246,
    -0.018709981814026833,
    0.042763907462358475,
    -0.09039932489395142,
    0.09996786713600159,
    -0.015323147177696228,
    -0.06260371953248978,
    0.1460253894329071,
    0.13356950879096985,
    0.014317121356725693,
    -0.02997150830924511,
    -0.0794777199625969,
    0.010234617628157139,
    0.05709412693977356,
    -0.044101227074861526,
    0.01887385919690132,
    0.034149061888456345,
    0.1482919454574585,
    -0.05490771681070328,
    0.014293570071458817,
    0.011256352998316288,
    -0.07384604960680008,
    -0.07562100887298584,
    0.007909595035016537,
    -0.057282425463199615,
    -0.054377663880586624,
    -0.030018610879778862,
    -0.008767150342464447,
    -0.007658748887479305,
    0.13149112462997437,
    -0.023679068312048912,
    0.02186751365661621,
    0.09421073645353317,
    -0.07069722563028336,
    0.03897340968251228,
    0.0047365641221404076,
    -0.016843540593981743,
    0.1021786704659462,
    0.22272728383541107,
    -0.04001608118414879,
    0.06574208289384842,
    0.028217565268278122,
    -0.08185591548681259,
    -0.043799612671136856,
    0.03905536234378815,
    -0.06420779973268509,
    0.03730490431189537,
    -0.061156563460826874,
    0.05924917012453079,
    -0.04711582139134407,
    -0.05842267721891403,
    -0.06460531055927277,
    -0.005869551096111536
  ]
}

Exact voice, prompts, model identifiers, source revision and usage records are preserved; credentials are excluded. These intended performance directions were withheld from the annotation models.

{
  "key": "de-00001",
  "language": "de",
  "language_tag": "de-DE",
  "voice": "Charon",
  "challenge_id": 1,
  "source": "existing_challenge_inspired",
  "title": "The Last Sunset on Our Beach",
  "instruction": "You're lying on the same beach where you fell in love thirty years ago, but this time your partner is dying beside you, their hand growing cold in yours. Begin with the sensory details—the warm sand, the waves—trying to hold onto peace, speaking softly about the beauty around you. But as you continue, let grief crack through your voice. The warmth of the sand becomes unbearable because they won't feel it tomorrow. Each wave that once brought peace now sounds like time running out. End with your voice breaking as you whisper that you're not ready, that no sunset has ever felt this heavy, transforming from attempted serenity into raw, vulnerable heartbreak.",
  "inspired_by": "Relaxing on a Sandy Beach",
  "prompt_style_key": "burst",
  "prompt_style_label": "Sparse vocal response",
  "script": {
    "spoken_text": "Schau mal, wie golden das Wasser glänzt... fast genau wie damals. Der Sand unter uns ist noch ganz warm. Aber deine Hand... <breath> sie wird so kühl. Jede Welle klingt plötzlich nur noch wie ein Herzschlag, der verstreicht. Wie Zeit, die uns wegrutscht. Ich will nicht, dass die Sonne untergeht. <sob> Bitte... ich bin noch überhaupt nicht bereit dafür.",
    "beat_styles": [
      "tender and quiet",
      "trembling realization",
      "fragile despair"
    ],
    "intended_emotions": [
      "Sadness",
      "Grief",
      "Love"
    ],
    "scene_note": "A dying evening on the beach turns from comforting reminiscence to raw heartbreak as the speaker realizes their partner's life is slipping away.",
    "language_check": {
      "method": "langdetect",
      "top_language": "de",
      "top_probability": 1.0
    },
    "inline_tags": [
      "breath",
      "sob"
    ],
    "word_count": 57
  },
  "script_request": {
    "systemInstruction": {
      "parts": [
        {
          "text": "You are a skilled dialogue writer for naturalistic, close-microphone voice acting.\nWrite a short first-person monologue that could be spoken in a real situation. The actor is\nspeaking to another person or themselves, not reciting a plot synopsis. Preserve the\nchallenge's specific situation and emotional development while avoiding melodrama,\ngeneric motivational language, exposition dumps, and perfect studio-script polish.\nHuman emotion is mixed, context-dependent, and gradually changing. Include small,\ncredible imperfections or micro-distractions only when they fit. Spoken text should\nhave about 58-68 words, normally four to six sentences, so that a natural performance\ntakes roughly 26-32 seconds. Use punctuation, ellipses, and occasional exclamation\npoints only where the thought calls for them. Never write stage directions in brackets\nor in the spoken text: Gemini 3.8 TTS will read all ordinary words aloud.\nUse only documented inline tags for momentary human vocal events. No music, foley,\nor background sound tags. Do not overuse vocal tags. Write the spoken dialogue ONLY in natural German (de-DE), not English.\nPreserve the challenge situation rather than translating it literally.\nKeep the inline vocal-event tags in English for TTS. Use 60-75 spoken German words. Keep profanity and graphic content out of the script.\n\nRespond ONLY as a JSON object with keys:\n  spoken_text: a single verbatim transcript including optional inline vocal tags;\n  beat_styles: an array of one to three brief English delivery directions, in order;\n  intended_emotions: one to three of the 40 EmoNet emotion labels if known, otherwise\n    concise ordinary emotion words;\n  scene_note: one sentence on why the vocal choice fits the challenge.\n"
        }
      ]
    },
    "contents": [
      {
        "role": "user",
        "parts": [
          {
            "text": "CHALLENGE TITLE: The Last Sunset on Our Beach\nCHALLENGE INSTRUCTION: You're lying on the same beach where you fell in love thirty years ago, but this time your partner is dying beside you, their hand growing cold in yours. Begin with the sensory details—the warm sand, the waves—trying to hold onto peace, speaking softly about the beauty around you. But as you continue, let grief crack through your voice. The warmth of the sand becomes unbearable because they won't feel it tomorrow. Each wave that once brought peace now sounds like time running out. End with your voice breaking as you whisper that you're not ready, that no sunset has ever felt this heavy, transforming from attempted serenity into raw, vulnerable heartbreak.\nPROMPT STYLE: Sparse vocal response\nSTYLE-SPECIFIC WRITING DIRECTION: Include one or two credible, situation-specific human non-speech vocal events as inline tags, such as <sigh>, <chuckle>, <gasp>, <groan>, <sob>, <grunt>, <breath>, or <short pause>. Only use an event that a real person in this scene might make; the scene still has coherent spoken language.\nDELIVERY BASELINE: naturalistic acting with one restrained, situation-appropriate vocal reaction; emotions flow without theatrical emphasis\nWrite all spoken words in natural German (de-DE). Make this a first-person monologue of roughly 30 seconds, with coherent speech and at most two restrained, situation-specific inline vocal-event tags. Choose human imperfections and flowing emotions only when they fit this exact challenge. Do not mention the studio voice or voice name in the dialogue."
          }
        ]
      }
    ],
    "generationConfig": {
      "responseMimeType": "application/json",
      "thinkingConfig": {
        "thinkingLevel": "low"
      },
      "temperature": 0.85
    }
  },
  "script_prompt": "CHALLENGE TITLE: The Last Sunset on Our Beach\nCHALLENGE INSTRUCTION: You're lying on the same beach where you fell in love thirty years ago, but this time your partner is dying beside you, their hand growing cold in yours. Begin with the sensory details—the warm sand, the waves—trying to hold onto peace, speaking softly about the beauty around you. But as you continue, let grief crack through your voice. The warmth of the sand becomes unbearable because they won't feel it tomorrow. Each wave that once brought peace now sounds like time running out. End with your voice breaking as you whisper that you're not ready, that no sunset has ever felt this heavy, transforming from attempted serenity into raw, vulnerable heartbreak.\nPROMPT STYLE: Sparse vocal response\nSTYLE-SPECIFIC WRITING DIRECTION: Include one or two credible, situation-specific human non-speech vocal events as inline tags, such as <sigh>, <chuckle>, <gasp>, <groan>, <sob>, <grunt>, <breath>, or <short pause>. Only use an event that a real person in this scene might make; the scene still has coherent spoken language.\nDELIVERY BASELINE: naturalistic acting with one restrained, situation-appropriate vocal reaction; emotions flow without theatrical emphasis\nWrite all spoken words in natural German (de-DE). Make this a first-person monologue of roughly 30 seconds, with coherent speech and at most two restrained, situation-specific inline vocal-event tags. Choose human imperfections and flowing emotions only when they fit this exact challenge. Do not mention the studio voice or voice name in the dialogue.",
  "script_system_prompt": "You are a skilled dialogue writer for naturalistic, close-microphone voice acting.\nWrite a short first-person monologue that could be spoken in a real situation. The actor is\nspeaking to another person or themselves, not reciting a plot synopsis. Preserve the\nchallenge's specific situation and emotional development while avoiding melodrama,\ngeneric motivational language, exposition dumps, and perfect studio-script polish.\nHuman emotion is mixed, context-dependent, and gradually changing. Include small,\ncredible imperfections or micro-distractions only when they fit. Spoken text should\nhave about 58-68 words, normally four to six sentences, so that a natural performance\ntakes roughly 26-32 seconds. Use punctuation, ellipses, and occasional exclamation\npoints only where the thought calls for them. Never write stage directions in brackets\nor in the spoken text: Gemini 3.8 TTS will read all ordinary words aloud.\nUse only documented inline tags for momentary human vocal events. No music, foley,\nor background sound tags. Do not overuse vocal tags. Write the spoken dialogue ONLY in natural German (de-DE), not English.\nPreserve the challenge situation rather than translating it literally.\nKeep the inline vocal-event tags in English for TTS. Use 60-75 spoken German words. Keep profanity and graphic content out of the script.\n\nRespond ONLY as a JSON object with keys:\n  spoken_text: a single verbatim transcript including optional inline vocal tags;\n  beat_styles: an array of one to three brief English delivery directions, in order;\n  intended_emotions: one to three of the 40 EmoNet emotion labels if known, otherwise\n    concise ordinary emotion words;\n  scene_note: one sentence on why the vocal choice fits the challenge.\n",
  "script_model": "gemini-3.8-flash",
  "script_usage": {
    "promptTokenCount": 720,
    "candidatesTokenCount": 197,
    "totalTokenCount": 917,
    "promptTokensDetails": [
      {
        "modality": "TEXT",
        "tokenCount": 720
      }
    ],
    "serviceTier": "SERVICE_TIER_STANDARD"
  },
  "tts_model": "gemini-3.8-flash-tts",
  "tts_request": {
    "contents": [
      {
        "role": "user",
        "parts": [
          {
            "text": "Schau mal, wie golden das Wasser glänzt... fast genau wie damals. Der Sand unter uns ist noch ganz warm. Aber deine Hand... <breath> sie wird so kühl. Jede Welle klingt plötzlich nur noch wie ein Herzschlag, der verstreicht. Wie Zeit, die uns wegrutscht. Ich will nicht, dass die Sonne untergeht. <sob> Bitte... ich bin noch überhaupt nicht bereit dafür.",
            "speech_metadata": {
              "style": "naturalistic acting with one restrained, situation-appropriate vocal reaction; emotions flow without theatrical emphasis; tender and quiet"
            }
          }
        ]
      }
    ],
    "generationConfig": {
      "responseModalities": [
        "AUDIO"
      ],
      "speechConfig": {
        "voiceConfig": {
          "voice": "Charon"
        }
      }
    }
  },
  "tts_usage": {
    "promptTokenCount": 91,
    "candidatesTokenCount": 842,
    "totalTokenCount": 933,
    "promptTokensDetails": [
      {
        "modality": "TEXT",
        "tokenCount": 91
      }
    ],
    "candidatesTokensDetails": [
      {
        "modality": "AUDIO",
        "tokenCount": 842
      }
    ],
    "serviceTier": "SERVICE_TIER_STANDARD"
  },
  "duration_s": 33.68,
  "audio_format": "mp3",
  "audio_bitrate_kbps": 128,
  "mp3_sha256": "3fdef6ec2a5e306994c4da3af5f0e7220b2804f49edf65b244d50c63493bb827",
  "voice_assignment_seed": 20261009,
  "voice_assignment": "balanced per language and source subset",
  "source_repo": "https://github.com/LAION-AI/Voice-Acting-Pipeline",
  "source_commit": "e3b0d5281558f7f73e1118c6a9b4c149bb8c0621",
  "source_file_sha256": "7ef759b639ef3646ca8a985664747e37ad93e4ab989f97dd8cece4f5201ceb6f"
}

{
  "humaneness_model": "laion/humaneness-ears-base-medium",
  "humaneness_revision": "818506970d93c809a3295a0a02c40dee8ff2bfc3",
  "caption_models": {
    "voice_repo": "laion/moss-voice-captioner",
    "voice_revision": "07cad84c33c676ea37979fb7ee844fd06ad92a76",
    "burst_repo": "laion/moss-vocal-burst-captioner",
    "burst_revision": "434e38f7e3eeaac0de4d3325b93a924e441153ce"
  },
  "original_inference_code_commit": "a6a286f",
  "procedural_format_code_commit": "dd1ac6e3b8bc2574b164680040b81703a1f3696f",
  "normalization_sha256": "c96503622619f63528a12ddcb28b19d4beddff7d94522d11b6a83dabad654f43",
  "original_pipeline_settings_hash": "1cbe629c2f1aef34160f2482e398fba26d131e90c582fff2f767158560a90916",
  "release_format_hash": "731fb1dee60a16c4a665fa556cacb5266d0aa303366b3741cd0a7b4b8298b8fc",
  "prompt_format_id": "humaneness-general-script-v2-20261011",
  "machine_generated_not_human_verified": true,
  "canonical_mixed_reference_z_and_caption_speech_reference_z_are_separate": true,
  "raw_source_archives_unchanged": true
}